Rail–local collaboration is vital for railway emergency response, yet plan texts are unstructured and difficult to analyze quantitatively. This study proposes an integrated framework combining large language model-based semantic extraction, dual-view knowledge graph construction, topology-based cascading failure simulation, and multidimensional resilience evaluation. Emergency plan texts are transformed into structured Triplet++ collaboration records and used to build two weighted networks: a railway-internal network and a full-domain rail–local network. Cascading failure is modeled through progressive loss of coordination nodes under multiple attack strategies. The Resilience Gain Coefficient assesses the relative value of cross-domain collaboration, while an entropy-weighted Composite Resilience Index measures absolute resilience. Experiments on seven railway emergency scenarios show that reasoning-guided extraction significantly improves the recovery of entities, relations, and cross-domain edges. Results reveal conditional resilience gains from rail–local collaboration: it enhances tolerance to random disturbances but may increase vulnerability to targeted attacks due to bridge-node dependence. Phase, profession, and hierarchy analyses further identify weak activation–response intervals, professional collaboration gaps, and inefficient command chains. These findings provide actionable structural evidence for improving cross-domain emergency plans.

Objective The rapid expansion of China's high-speed railway network has imposed increasingly stringent requirements on environmental safety protection. Foreign object intrusion into the railway clearance zone poses a serious threat to operational safety, particularly in high-risk areas such as tunnels, cuttings and mountainous sections that are prone to rockfalls, debris flows and landslides. Existing single-sensor monitoring systems often suffer from high false alarm and missed detection rates under adverse weather or illumination conditions, limiting their reliability. To address this limitation, this study proposes a geometric-prior-driven multi-sensor fusion method that integrates LiDAR and camera data. The method leverages the geometric constraints of the railway clearance envelope to improve robustness, reduce false alarms and achieve zero missed detection of small obstacles in complex environments. Methods The proposed system adopts an "independent perception + decision-level fusion" architecture. On the LiDAR side, a digital ground height model is constructed through spatial partitioning and dynamic modeling, enabling adaptive terrain estimation across complex track environments. Obstacle candidates are extracted by calculating relative height differences with respect to the ground model and applying density-based clustering. To mitigate severe weather interference, a multi-feature noise filtering mechanism is introduced, incorporating reflection intensity, echo patterns, spatial neighborhood consistency and temporal stability. On the camera side, YOLOv8 is employed as the two-dimensional (2D) detection network to identify targets, with improved small-object recall due to its anchor-free design and efficient backbone architecture. Data fusion is guided by geometric priors: three-dimensional (3D) LiDAR candidates are projected onto the image plane and matched with 2D bounding boxes based on intersection-over-union and spatial consistency. The matched candidates are then processed by a lightweight neural network (approximately 2.4 & times;10 & sup3; parameters) that outputs fusion confidence scores. A thresholding step combined with non-maximum suppression generates the final detection results. To ensure real-time performance, the entire pipeline is accelerated using GPU-based KD-tree and octree indexing together with CUDA kernels, significantly reducing computational overhead. Results and Discussion Extensive validation was conducted through both laboratory and field experiments across multiple railway environments. Six rounds of large-scale experiments were carried out at the National Railway Track Test Center, the Xinjin testing base and operational sections of the Chongqing-Kunming and Guangzhou-Shenzhen-Hong Kong high-speed railways. The experimental conditions included clear weather, daytime and nighttime illumination, heavy rainfall up to 32 mm/h, strong light interference exceeding 20,000 lx, fog (visibility 500 m), thick fog (200 m) and dense fog (50 m). The results indicate that under clear and rainy conditions, the system achieved zero missed detections for obstacles with an edge length as small as 20 cm within a monitoring range of 70 m along the forward and lateral clearance boundaries. Even under heavy rainfall and strong light interference, the missed detection rate remained negligible, while the false alarm frequency averaged only 0.78 occurrences per day per monitoring site. Compared with single-sensor baseline systems, the geometric-prior-driven fusion approach reduced false alarms by more than 60% and improved mean average precision (mAP) by 4-6 percentage points. In fog conditions, performance degradation was observed due to laser diffraction effects; however, the fusion strategy still outperformed single-modality detectors by maintaining stable recall through complementary visual information. The geometric prior plays a crucial role by restricting detection within the clearance envelope, effectively eliminating irrelevant candidates and guiding cross-modal matching. Consequently, only geometrically consistent targets are retained, improving both detection accuracy and computational efficiency. Furthermore, the lightweight neural network is suitable for real-time edge deployment, achieving an end-to-end processing latency of less than 130 ms on embedded devices with 16 TOPS computing capability. Conclusions This study presents a geometric-prior-driven LiDAR-camera fusion method for monitoring foreign object intrusion in railway environments. By incorporating clearance constraints and track geometry into both LiDAR-based obstacle detection and cross-modal fusion, the proposed approach demonstrates strong adaptability to complex terrain and robustness under adverse weather conditions. Field experiments confirm that the method achieves zero missed detections for small obstacles while maintaining a low false alarm rate, thereby meeting the stringent safety requirements of railway monitoring systems. Future work will extend the framework by integrating heterogeneous sensors such as four-dimensional millimeter-wave radar to further enhance detection performance in dense fog and snowfall. Additional experiments will also be conducted in plateau and mountainous regions to improve environmental adaptability and strengthen the reliability and scalability of railway safety monitoring systems.

Constructing a static exterior model of railway passenger station is a preliminary and crucial step in achieving a digital twin station. Station modeling often relies on manual techniques, requiring significant labor for stations spanning tens of thousands of square meters. While the interior decoration is symmetrical, the asymmetrical placement of equipment often leads to confusion in manual modeling. This paper proposes a novel station static model reconstruction method - MSCRAGS (Mobile vehicle-Sparse Sampling-Colmap-Resolution adjustment-Gaussian Splatting). The MSCRAGS integrates the requirements for flexible control of static exterior models by utilizing mobile vehicle for data collection and incorporates the sparse multi-view spatial sampling approach. It involves collecting multi-height and multi-angle appearance color data of passenger operation elements to construct preliminary point clouds. Subsequently, 3d Gaussian splatting is employed for rendering, achieving high-fidelity reconstruction of production elements. Moreover, according to the requirements for the precision control of static exterior models, the rendering effects are reshaped with resolution adjustment to obtain static exterior models of station production elements at various resolutions. Experiments conducted at Qinghe Station demonstrate that compared to other state-of-art modeling methods, our approach significantly reduces modeling time and improves modeling accuracy, showing superior performance in modeling high-fidelity indices.

Purpose The paper aims to solve the problem of personnel intrusion identification within the limits of high-speed railways. It adopts the fusion method of millimeter wave radar and camera to improve the accuracy of object recognition in dark and harsh weather conditions. Design/methodology/approach This paper adopts the fusion strategy of radar and camera linkage to achieve focus amplification of long-distance targets and solves the problem of low illumination by laser light filling of the focus point. In order to improve the recognition effect, this paper adopts the YOLOv8 algorithm for multi-scale target recognition. In addition, for the image distortion caused by bad weather, this paper proposes a linkage and tracking fusion strategy to output the correct alarm results. Findings Simulated intrusion tests show that the proposed method can effectively detect human intrusion within 0–200 m during the day and night in sunny weather and can achieve more than 80% recognition accuracy for extreme severe weather conditions. Originality/value (1) The authors propose a personnel intrusion monitoring scheme based on the fusion of millimeter wave radar and camera, achieving all-weather intrusion monitoring; (2) The authors propose a new multi-level fusion algorithm based on linkage and tracking to achieve intrusion target monitoring under adverse weather conditions; (3) The authors have conducted a large number of innovative simulation experiments to verify the effectiveness of the method proposed in this article.

PurposeThe safety of high-speed rail operation environments is an important guarantee for the safe operation of high-speed rail. The operating environment of the high-speed rail is complex, and the main factors affecting the safety of high-speed rail operating environment include meteorological disasters, perimeter intrusion and external environmental hazards. The purpose of the paper is to elaborate on the current research status and team research progress on the perception of safety situation in high-speed rail operation environment and to propose directions for further research in the future.Design/methodology/approachIn terms of the mechanism and spatio-temporal evolution law of the main influencing factors on the safety of high-speed rail operation environments, the research status is elaborated, and the latest research progress and achievements of the team are introduced. This paper elaborates on the research status and introduces the latest research progress and achievements of the team in terms of meteorological, perimeter and external environmental situation perception methods for high-speed rail operation.FindingsBased on the technical route of “situational awareness evaluation warning active control,” a technical system for monitoring the safety of high-speed train operation environments has been formed. Relevant theoretical and technical research and application have been carried out around the impact of meteorological disasters, perimeter intrusion and the external environment on high-speed rail safety. These works strongly support the improvement of China’s railway environmental safety guarantee technology.Originality/valueWith the operation of CR450 high-speed trains with a speed of 400 km per hour and the application of high-speed train autonomous driving technology in the future, new and higher requirements have been put forward for the safety of high-speed rail operation environments. The following five aspects of work are urgently needed: (1) Research the single factor disaster mechanism of wind, rain, snow, lightning, etc. for high-speed railways with a speed of 400 kms per hour, and based on this, study the evolution characteristics of multiple safety factors and the correlation between the high-speed driving safety environment, revealing the coupling disaster mechanism of multiple influencing factors; (2) Research covers multi-source data fusion methods and associated features such as disaster monitoring data, meteorological information, route characteristics and terrain and landforms, studying the spatio-temporal evolution laws of meteorological disasters, perimeter intrusions and external environmental hazards; (3) In terms of meteorological disaster situation awareness, research high-precision prediction methods for meteorological information time series along high-speed rail lines and study the realization of small-scale real-time dynamic and accurate prediction of meteorological disasters along high-speed rail lines; (4) In terms of perimeter intrusion, research a multi-modal fusion perception method for typical scenarios of high-speed rail operation in all time, all weather and all coverage and combine artificial intelligence technology to achieve comprehensive and accurate perception of perimeter security risks along the high-speed rail line and (5) In terms of external environment, based on the existing general network framework for change detection, we will carry out research on change detection and algorithms in the surrounding environment of high-speed rail.

In recent years, the safety issues of high-speed railways have remained severe. The intrusion of personnel or obstacles into the perimeter has often occurred in the past, causing derailment or parking, especially in the case of bad weather such as fog, haze, rain, etc. According to previous research, it is difficult for a single sensor to meet the application needs of all scenario, all weather, and all time domains. Due to the complementary advantages of multi-sensor data such as images and point clouds, multi-sensor fusion detection technology for high-speed railway perimeter intrusion is becoming a research hotspot. To the best of our knowledge, there has been no review of research on multi-sensor fusion detection technology for high-speed railway perimeter intrusion. To make up for this deficiency and stimulate future research, this article first analyzes the situation of high-speed railway technical defense measures and summarizes the research status of single sensor detection. Secondly, based on the analysis of typical intrusion scenarios in high-speed railways, we introduce the research status of multi-sensor data fusion detection algorithms and data. Then, we discuss risk assessment of railway safety. Finally, the trends and challenges of multi-sensor fusion detection algorithms in the railway field are discussed. This provides effective theoretical support and technical guidance for high-speed rail perimeter intrusion monitoring.

Due to adjustments to the operation plan of guided trains at high-speed railway stations, a large amount of information is inevitably displayed, sometimes with delays, omissions, and misalignments. The effective management of guidance information can provide important support for the personnel flow operation of high-speed railway stations. Aiming to meet the requirements of high real-time and high accuracy of guided job control, a closed-loop control method based on a guided job is proposed, which provides enhanced text detection and recognition in a target area. Firstly, using the introduction of the triplet attention mechanism in YOLOv5 and the addition of fusion modules, the feature pyramid network is used to enhance the effective feature and feature interactions between the modules to improve the detection speed of the display. Then, the text on the guide screen is recognized and extracted in combination with the PaddleOCR model, and then, the results are proofread against the original plan to adjust the screen information. Finally, the effectiveness and feasibility of the method are verified by experimental data, with the accuracy of the improved model reaching 90.6% and the speed reaching 1 ms, which meets the requirement of real-time closed-loop control of Screen-Based Guidance Operations.

Station daily operation management is an essential task in the railway industry. To enhance the efficiency of this process, we propose a digital twins four-dimensional model for daily station operations (DTSDO) in this study. The DTSDO model consists of four distinct components: Passenger Station Physical Entity (PSPE), Passenger Station Virtual Entity (PSVE), Digital Twins Connections (DTCS), and Daily Operation Twins Services (DOTS). The DTSDO is driven by the fundamental data of passengers and trains, and it uses the eEPCD to link and control the Six essential production elements within the station based on a time sequence. The communication between the PSPE and PSVE is established via the HTTP protocol. The DOTS accurately portrays the virtual representation of the reality passenger station. Furthermore, we have validated the DTSDO model using Qinghe station as a specific case study. We created a one-to-one model of the station and tested its comprehensive functionalities.

In order to improve the supervision management efficiency on railway safety, several railway safety supervision trajectories are computed and calculated by the multiple objectives PSO algorithm. Firstly, the objective functions, which are mainly composed of the whole distance and the sum square error on solving safety problem time, are introduced to deeply understand the safety supervision trajectory problem. Secondly, the decision variables in each particle are the sorted order according to the optimized parameters and some main steps on the multiple objectives PSO algorithm are also provided to handle with the corresponding problem. Thirdly, numerical results highlight that the multiple objectives PSO algorithm can provide four optimal trajectories for four safety supervisors and it is key to analyze the Pareto front and the Pareto solutions during the whole evolutionary process.

本文旨在应对高铁周界环境复杂、小目标多等情况,研究周界入侵行为的识别与跟踪问题,并提出一种改进ByteTrack算法.本文融合YOLOv7-X与BYTE数据关联方法对模型进行改进,并且引入卷积块注意力机制以提升周界复杂环境下前景目标的识别效果,利用空间-深度转化模块优化跨步卷积与池化层,改善小目标识别时下采样导致的细粒度信息丢失情况.制作铁路周界入侵数据集进行实验,实验结果表明,改进后的模型平均精度达到 95.6％,提升了 9.4％,对大中小目标识别的平均精度均有提升,尤其是对小目标识别效果提升显著,提升了 22.2％.结果表明改进ByteTrack算法在高铁周界复杂环境下能实现入侵行为的识别与跟踪,为高铁周界防护提供技术支持.

针对铁路综合监控视频中不同远近行人成像面积差异较大、自然环境变化产生干扰等因素造成的检测难题,提出一种改进FairMOT框架的周界入侵检测方法.首先,针对监控视频中不同远近的行人,通过在FairMOT框架中引入感受野模块,丰富不同成像大小行人检测所需的感受野,以更好地提取不同尺度特征信息;其次,针对夜晚时段方法检测性能较低的问题,在编码解码网络后融合空间注意力模块,强化夜间前景行人关键特征,同时优化目标跟踪和判断流程,实现稳定检测;然后,针对缺乏大量学习样本的问题,使用行人检测跟踪数据集与铁路真实数据集混合增强训练,提高方法在全天候检测中的泛化性和鲁棒性;最后,在MOT17数据集和铁路真实数据集上,对改进FairMOT检测方法与CenterTrack,Bytetrack等方法进行对比试验.结果表明:提出的改进FairMOT检测方法在白天和夜晚对不同大小目标检测中,均取得了最高的准确率和召回率调和均值,检测性能最好;方法检测速率为25.2帧·s-1,能够满足实时检测要求.改进的FairMOT检测方法可以更有效地应用于实际铁路周界入侵检测场景.

人工智能日益融入经济发展各个领域,并对各行各业的运行方式产生深远影响.伴随着人工智能大模型的不断发展与应用,对数据、算法、算力等方面均提出了更高的要求.目前,人工智能已在铁路行业多个领域发挥重要作用,但铁路人工智能应用在平台能力、技术体系和业务应用建设中还存在一些问题.文章分析铁路人工智能的应用现状及需求,提出铁路人工智能平台的建设目标,设计铁路人工智能平台的总体架构及功能,研究铁路人工智能平台关键技术,对促进人工智能在铁路行业的系统应用具有积极意义.

Platform end security control is of utmost importance in railway passenger station safety management. Station personnel track intruders using a single-line LiDAR. This paper analyzes the characteristics of intruder targets at the platform end and the point cloud features of intruders collected by the single-line LiDAR. An algorithm for tracking intruder targets at the platform end of the single-line LiDAR point cloud is proposed in this paper. By integrating spatial and temporal factors, spatial data processing is conducted in the first stage to process the single-frame data and construct multi-arc rectangles of intruder targets. In the second stage, data correlation calculations are performed on the radar frames based on the time series, resulting in a correlation matrix. The Hungarian algorithm is employed for multi-object correlation between frames. Specifically, for one-to-many matching situations, the dynamic time warping method is used to associate and match the arcs within the multi-arc rectangles, enabling secondary matching for inter-frame target alignment and achieving intruder target tracking at the platform end using the single-line LiDAR point cloud. Experimental results demonstrate that the proposed algorithm can track platform intruder targets.

Deep learning techniques continue to be used in various applications in recent years. However, when it is difficult to obtain adequate training samples, the performance of the depth model will degrade. Although few-shot learning and data enhancement techniques can relieve this dilemma, the diversity of real data is too large to simulate. To tackle this challenge, we study a novel method, Data Augmentation Scheme For Few-Shot Object Detection (DA-FSOD), to improve the efficiency of model training on visual tasks. Specifically, to expand data augmentation space, we build a data augmentation operation pool (DAOP) based on several common-applied image process operations. Then we propose a novel data augmentation scheme, the series and parallel connection scheme, which superimposes the effects of different operations to generate diverse variants. To further explore and utilize the deep feature information, we leverage the semantic information of input image in model and propose imposed semantic data augmentation which augments training set semantically via deep features of augmented variants. The proposed method successfully enhanced the model performance. We validated our approach using extensive experiments on the domain of few-shot object detection. The results showed remarkable gains compared to state-of-the-art methods.

为提高铁路客站的服务质量,通过详细分析显示技术从有屏到无屏显示的三个发展阶段,得出当前技术正处于有屏到无屏的过渡阶段,进一步提出无屏显示技术内涵及特征,研究无屏显示技术的四类成像原理及其对应的显示技术.根据每种无屏显示技术的研究及应用的情况,从铁路客站的实际需求出发,针对实际业务开展无屏显示技术的前瞻性应用研究,提出每种显示技术的应用场景及建议,为铁路客站的智能化发展提供新的思路.

在《国家综合立体交通网规划纲要》明确提出"推动干线铁路、城际铁路、市域(郊)铁路融合建设,并做好与城市轨道交通衔接协调,构建运营管理和服务'一张网,,实现设施互联、票制互通、安检互认、信息共享、支付兼容"以来,我国智能高速铁路、智慧地铁建设取得显著成效.城际铁路作为连接干线铁路和城市轨道交通的通道,积极开展其智能化总体设计,对于推动轨道交通四网高质量融合具有极其重要的意义.系统分析城际铁路智能化总体需求,阐述智能城际铁路的发展目标及内涵定义,采用体系工程设计方法,提出智能城际铁路的业务应用架构、数据架构和技术架构.该研究可为城际铁路智能化建设提供顶层设计建议.

京张高速铁路是支撑北京冬奥会成功举办的重要交通基础设施.为进一步提升冬奥赛事期间京张高速铁路旅客服务的智能化水平,构建了面向冬奥的京张高速铁路旅客服务智能化技术体系架构,深入研究了京张高速铁路旅客服务智能化成套关键技术,涵盖站车智能旅客服务环境与生产要素泛在感知、客运集中管控平台智能提升、基于5G的列车多媒体奥运信息服务、旅客全行程智慧信息服务、人员异常行为感知与预警、站车应急处置决策与协同联动、站台门自适应控制等技术.相关研究成果已在京张高速铁路完成应用示范,满足了冬奥赛事期间国际化旅客出行和车站生产运营的需求,有效提升了旅客出行服务的智能化水平,提高了车站生产管理效率和安全保障能力,保障了京张高速铁路稳定、有序运行.

针对隧道工程地质及环境条件复杂,不可预见安全风险因素多,质量控制和环水保要求高,施工组织协调任务重,研究采用数字信息化技术实现隧道工程安全优质高效施工建设.首先,研究提出基于BIM的隧道施组进度跨任务综合预警方法和基于BIM的三维可视化进度精细化管理;其次,研究基于数字视频分析、短距离人机定位技术实现隧道内人员和机械设备安全定位管理,基于BIM、物联网、物探法、钻探法、自动监测等隧道施工不良地质超前探测预报预警、围岩变形动态监控量测、安全步距动态分析监控、有害气体监测预警等隧道工程施工多维度安全风险监控;再次,研究BIM、三维扫描、移动互联等新技术与隧道施工开挖业务融合,实现对隧道工程施工超欠挖和平整度、衬砌防脱空以及工程实体质量专业分析和自动预警等;最后,结合实际工程应用验证关键技术和应用效果,为隧道工程施工建设数字信息化管理提供借鉴参考.

为实现铁路节能降耗,助力实现"碳中和",围绕铁路客站列车到发信息、区域环境照度等影响因素开展铁路客站照明节能控制策略研究.首先,基于列车到发时间、旅客到站人数等信息生成照明控制计划,并对站台进行合理区域划分,采用自适应加权融合法测量各子区域照度;其次,结合照明控制计划与区域照度,设计照明节能控制规则,研究提出节能控制策略,通过对比3种不同区域划分工况下的能耗统计,以节能效果更佳的区域划分进行设计;最后,选取某高速铁路客站站台进行实验,择优选用4回路工况的节能控制策略,并与长亮模式、既有节能控制模式相比,验证了在保障站台运营安全基础上,提出的节能控制策略可以有效降低客站电能消耗.

为解决卫星观测视野不良或导航信号受干扰时,北斗定位解算失效导致列车失去定位,提出基于克里金法的场强定位方法对卫星导航定位方式进行有效替代.利用列车沿线性轨道固定往复运行且信号交织覆盖的特性,通过克里金法对采集的无线场强进行插值,能够在降低无线数据采集工作量的同时实现无线场强数据库快速建立和动态更新.通过Tower Collector软件,同步采集GNSS定位数据和基站场强数据,利用克里金法插值后的无线场强数据库对终端位置进行实时解算.结果表明,基于克里金法的场强定位精度与GNSS定位的平均误差为70.5 m,最大误差243.5 m,可在卫星定位失效时,有效地为列车提供参考定位,该方法可行性较高.